Tweak Lambertian reflection description

My main object was to address some confusing wording of the last
sentence of the first paragraph, but I also massed other parts of this
paragraph.

Author: hollasch

alpha/books/RayTracingInOneWeekend.html

@@ -2452,13 +2452,13 @@
 ---------------------------
 Using a diffuse model that scatters vectors evenly about the hemisphere produces a nice and soft
 diffuse model, but we can definitely do better. A more accurate representation of real diffuse
-objects is the _Lambertian_ distribution, which has a scattering distribution that is propertional
-to $\cos (\phi)$, where $\phi$ is the angle from the normal. This means that a ray is most likely to
-scatter toward the normal and decreases in likelihood as the direction becomes more tangential. This
-distribution is notably no longer uniform, but the _Lambertian_ distribution does a better job of
-modeling objects in the real world than our previous uniform scattering. We can create this
-distribution by picking random points on the surface of the unit sphere, and then offsetting them
-along the surface normal.
+objects is the _Lambertian_ distribution, which scatters reflected rays in a manner that is
+propertional to $\cos (\phi)$, where $\phi$ is the angle between the reflected ray and the surface
+normal. This means that a ray is most likely to scatter in directions close to the normal, and less
+likely to scatter in directions more tangent to the surface. This distribution is notably no longer
+uniform, but the _Lambertian_ distribution does a better job of modeling objects in the real world
+than our previous uniform scattering. We can create this distribution by adding a random unit vector
+to the normal vector.
 
 At the point of intersection on a surface there is the hit point, $\mathbf{p}$, and there is the
 normal of the surface, $\mathbf{n}$. At the point of intersection, this surface has exactly two